Session Type: Poster Session (Tuesday)
Session Time: 9:00AM-11:00AM
Background/Purpose: Interstitial lung disease (ILD) is an increasingly important problem for patients with Rheumatoid Arthritis (RA). However, current approaches to ILD case finding are suboptimal or have been evaluated only in limited settings. Our objective was to develop, refine, and validate a claims-based algorithm to identify both prevalent and incident ILD in a cohort of RA patients enrolled in Medicare compared to the gold standard of medical record review.
Methods: We used Medicare administrative claims data 2006-2015 to derive a cohort of RA patients and then identified suspected ILD using variations on several features of the data. These included various ILD diagnosis codes using ICD9/10 coding, evaluation by pulmonologists and rheumatologists, occurrence of chest CT, lung biopsy results, and pulmonary function tests. Suspected ILD cases were evaluated at five participating medical centers: Duke, Medical University of South Carolina, University of Alabama at Birmingham, University of North Carolina, and Vanderbilt. The centers identified patients in their medical systems using either a search tool run against a central data warehouse or repository (e.g. i2b2) or a local ILD registry. Medical record reviewers abstracted the clinical data, including physician notes, CT scan and imaging results, lung pathology reports, and pulmonary Function tests (PFTs), into a case report form which was then adjudicated by two ILD experts (pulmonology and rheumatology). Discordance in adjudication was resolved by consensus. The positive predictive value (PPV), the primary outcome of the study, was calculated for each ILD algorithm for both prevalent and incident ILD and 95% confidence interval (CI) using a binomial distribution.
Results: We identified 264 linkable RA patients with sufficient data to evaluate for ILD. Overall, 115 (44.6%) of suspected cases (based on a highly sensitive case finding approach) were classified as ILD. The most common diagnosis codes identified in this initial search were 515 (Postinflammatory pulmonary fibrosis, 33%), 518.89 (Other disorders of lung, 24%), 714.81 (rheumatoid lung, 9%), 793.19 (Other nonspecific abnormal finding of lung field, 8%), and J84 (Other interstitial pulmonary diseases with fibrosis, 6%). 516 (Other alveolar and parietoalveolar pneumonopathy, 3%). A total of 36.0% of cases were hospitalized (4.9% primary diagnosis code, 31.1% non-primary diagnosis code), and the remainder were outpatient. The best performing algorithm for prevalent ILD had a PPV of 77% (95% CI 67%-85%) and for incident ILD was 74.2% (95% CI 58.9-89.6%).
Conclusion: ILD case finding in RA patients using administrative claims data is feasible and has reasonable accuracy. Both prevalent and incident ILD can be identified.
To cite this abstract in AMA style:Guthrie M, Shah A, Lobo L, Oates J, Clinton C, Annapureddy N, Xie F, England B, Curtis J. Validation of a Claims Algorithm to Identify Interstitial Lung Disease in Patients with Rheumatoid Arthritis [abstract]. Arthritis Rheumatol. 2019; 71 (suppl 10). https://acrabstracts.org/abstract/validation-of-a-claims-algorithm-to-identify-interstitial-lung-disease-in-patients-with-rheumatoid-arthritis/. Accessed April 11, 2021.
« Back to 2019 ACR/ARP Annual Meeting
ACR Meeting Abstracts - https://acrabstracts.org/abstract/validation-of-a-claims-algorithm-to-identify-interstitial-lung-disease-in-patients-with-rheumatoid-arthritis/